Abstract
Regression models were used to predict whole-body fat percentage from as few as 13 cross-sectional magnetic resonance images from each of 12 pigs. Models were fit indirectly, by relating fat percentage to statistics averaged over all images from each pig, and relating these to individual images. Predictors were checked for robustness against errors in image locations.
Original language | English |
---|---|
Pages (from-to) | 179-184 |
Number of pages | 6 |
Journal | Journal of the Royal Statistical Society. Series D (The Statistician) |
Volume | 41 |
Issue number | 2 |
Publication status | Published - 1992 |